基于即时学习的软测量建模实时性改进

摘要

Real‐time performance is an important issue of the Just‐In‐Time‐Learning based method .A re‐al‐time performance improvement strategy is proposed to enhance the online modeling efficiency of JITL based soft sensor by controlling the frequency of updating local models .First ,similarity factor could be calculated to decide whether or not to update the local soft sensor model ,compared to a predefined cutoff value .If the new data sample is considered to be very similar to that of the corresponding operation re‐gion ,the local model updating is not needed .Otherwise ,the local model should be re‐constructed .The superiority of the proposed method over the conventional methods is demonstrated through a case study of a Debutanizer column process .%针对传统即时学习软测量建模过程中实时性改进不足的问题，提出了一种自动降低局部模型更新频率的即时学习实时性改进方法。通过相似性测度衡量待测输入样本是否变化，以检测系统的工作点是否发生变化。如果系统工作点发生变化，则更新局部模型后预测输出，并更新相似性测度比对阈值；否则判定系统工作于稳态，无需更新局部模型。实验结果验证了提出的实时性改进算法既能够显著提高即时学习的建模实时性，又能够保证较高的预测准确性。